A review of prosthetic interface stress investigationsmental investigations is difficult, as the...
Transcript of A review of prosthetic interface stress investigationsmental investigations is difficult, as the...
WDepartment ofVeterans Affairs
Journal of Rehabilitation Research andDevelopment Vol . 33 No . 3, July 1996Pages 253-266
A review of prosthetic interface stress investigations
M. Barbara Silver-Thorn, PhD; John W. Steege, MSME; Dudley S. Childress, PhDMarquette University, Department of Biomedical Engineering, Milwaukee, WI 53201-1881 ; NorthwesternUniversity, Prosthetics Research Laboratory, Chicago, IL 60611-4496
Abstract—Over the last decade, numerous experimental andnumerical analyses have been conducted to investigate thestress distribution between the residual limb and prostheticsocket of persons with lower limb amputation . The objectivesof these analyses have been to improve our understanding ofthe residual limb/prosthetic socket system, to evaluate theinfluence of prosthetic design parameters and alignmentvariations on the interface stress distribution, and to evaluateprosthetic fit . The purpose of this paper is to summarize theseexperimental investigations and identify associated limita-tions . In addition, this paper presents an overview of variouscomputer models used to investigate the residual limbinterface, and discusses the differences and potential ramifica-tions of the various modeling formulations . Finally, thepotential and future applications of these experimental andnumerical analyses in prosthetic design are presented.
Key words : amputation, finite element analysis, pressure,prosthetics, stress.
INTRODUCTION
A prosthesis is often used to restore appearanceand functional mobility to individuals following lowerlimb amputation . Coupling between the residual limband the prosthesis is typically achieved by a socket,
This material is based upon work supported by the Department ofVeteran Affairs, Rehabilitation Research and Development Service,Washington, DC, and the National Institute of Disability and Rehabilita-tion Research.All terms in italics and followed by an asterisk ( 5 ) are defined in theGlossary at the end of this article.
Address all correspondence and requests for reprints to : M. BarbaraSilver-Thom, PhD, Marquette University, Department of Biomedical Engi-neering, Olin Building, Room 301, P .O . Box 1881, Milwaukee, WI53201- 188 1 ; email: barbara@silver .bien .mu .edu
which surrounds the residual limb, and to which theremaining components of the prosthesis are attached . Incurrent prosthetic practice, the socket is a criticalelement of a successful prosthesis, as it is the solemeans of load transfer between the prosthesis and theresidual limb.
The soft tissues of the residual limb are notwell-suited for load bearing, and the load tolerance ofthe residual limb depends on the biological andphysiological structure of these tissues . Whenevertissues are exposed to excessive or prolonged loading,there is a risk of tissue trauma due to local circulatorydeficits and/or abrasion . Thus, for persons with lowerlimb amputation, where large loads must be borne bythe soft tissues, great care is taken in the design of theprosthetic socket to minimize discomfort and possibletissue trauma. Other areas of rehabilitation where softtissues are exposed to load are subject to similarproblems (seat cushion and mattress design for wheel-chair and bedridden subjects, orthotic braces,orthopaedic walking casts, and pedorthics for individu-als with insensate feet, for example) . Nevertheless,although a review paper on interface stress has rel-
evance to many areas, this paper will focus on softtissue/support systems for persons with amputation.
One socket design that has shown success inbalancing the physiological and the load-bearing factorsfor persons with below-knee (BK) amputation is thepatellar-tendon-bearing (PTB) socket, initially devel-oped at the University of California at Berkeley in the
late 1950s . The basic concept of the PTB prosthetic
socket is to distribute the load over areas of the residuallimb in proportion to the ability of the tissues to tolerate
253
254
Journal of Rehabilitation Research and Development Vol . 33 No. 3 1996
load. Load is borne primarily on the patellar tendon(hence the socket's name), medial and lateral flares ofthe tibia, and popliteal area . The socket pre-compressesthe tissues of the residual limb in these load areas sothat forces may be preferentially distributed and themovement of the socket relative to the skeleton isminimized. The PTB socket is thus not a replica of theresidual limb, but instead includes appropriate shapemodifications (rectifications) so that pressure tolerantareas bear the majority of the load and pressuresensitive areas are relieved of load . These shapemodifications vary for each person with amputation andeach prosthesis, due to differences in residual limbgeometry, tissue stiffness, and load tolerances of thetissues.
The fitting of a prosthesis is an empirical process.The prosthetist has no quantitative information regard-ing the load distribution of the soft tissues and must relyon what he/she has been taught and/or has experienced,feedback from the patient, and indirect indications ofload (i.e ., skin blanching) to gage socket fit . Knowledgeof the interface stress distribution between the residuallimb and the prosthetic socket would enable objectiveevaluation of prosthetic fit, and might advance pros-thetic socket design through direct application ofscience and engineering.
It is this desire for quantitative description ofprosthetic interface stress distribution that has motivatedthe experimental and numerical investigations summa-rized in this paper. The objective of this paper is toprovide a summary of past work, to discuss thelimitations and potential sources of error of suchinvestigations, and to speculate on work made possiblethrough advances in pressure/force transducer designsand/or computer hardware, computational methods, andcomputer software.
Experimental AnalysesSeveral groups have investigated the stress distri-
bution between the residual limb and prosthetic socketfor both persons with BK and above-knee (AK)amputation in laboratory and/or clinical settings . Vari-ous means of pressure measurement have been used toinvestigate the effects of prosthetic alignment, relativeweight-bearing, muscle contraction, socket liners, andsuspension mechanisms on the interface pressure distri-bution (Table 1, 1-27) . Most experimental stressmeasurements have been limited to specific sites aroundthe limb, as measurements can only be obtained attransducer locations . Note that for the purposes of this
paper, pressure will be equated with normal stress, andshear will be equated with the tangential stresses.Therefore, complete description of the interface stressdistribution includes both pressure and shear.
Direct comparison of the results of these experi-mental investigations is difficult, as the interfacepressure measures are highly dependent on both themeans of measurement (i.e ., type of transducer) and thecalibration method employed . The transducers devel-oped by Sanders (20) and Williams et al . (6) are theonly transducers that enable simultaneous measurementof pressure and shear stresses . Ferguson-Pell hasreviewed several key factors in transducer use andselection relevant to stress measures between the softtissues of the body and a support structure (28,29).
The majority of the interface pressure measuringtechniques summarized in Table 1 require fabrication ofspecial sockets . These techniques have limitations inthat the mounting of the pressure transducers in thesocket requires that ports be tapped/drilled into thesocket wall, permanently altering the prosthesis ; thetransducers are expensive ; and the interface pressuresare obtained for only a relatively small portion of theinterface. In addition, many pneumatic transducers donot have sufficiently quick response to enable pressuremeasurement during dynamic loading (i .e ., gait) . Hy-draulic sensors may be temperature sensitive anddifficult to calibrate . The calibration of all of thesetransducers is critical to data analysis and interpretation.Transducers designed for hydrostatic/pneumatic applica-tions, but subjected to mechanical load, need to becalibrated appropriately (12,13) . Finally, because of thefinite thickness of the transducer and/or stiffnessincompatibility of the transducer and the interfacematerials, transducers placed between the residual limband the prosthetic socket may create stress concentra-tions that result in measurement anomalies.
Recent commercial developments include systemsto measure interface stresses for both seating andprosthetic systems. These systems were designed forroutine clinical application, and thus do not requiremodification of the seat cushion and/or prostheticsocket.
Tekscan, Inc . (Boston, MA) markets several bio-medical pressure measurement systems, including sys-tems to analyze tooth pressure (dental occlusion), footpressure (30), seat pressure, and hospital bed pressures.Each of these systems utilizes a grid-based sensor inwhich the rows and columns are separated by a polymerwhose electrical resistance varies with force . These
255
SILVER-THORN et al . Review of Prosthetic Stress Investigations
Table 1.Summary of experimental residual lower-limb interface pressure studies reported in the
literature.
Investigation
Transducer type
Reference
quantification of
dye-impregnated socksprosthetic fit
flexible force transducerhydraulic transducerhydraulic transducerdiaphragm transducer array
prosthetic liner effects
diaphragm transducer array
alignment variation effects
beam transducerdiaphragm transducerdiaphragm transducer
load state variations
diaphragm transducer
effects of muscle contraction
diaphragm transducer
suspension effects
diaphragm transducerdiaphragm transducer
finite element model
diaphragm transducervalidation
diaphragm/beam transducer
(1) Meier et al.(1) Meier et al.(2) VanPijkeren et al.(3) Isherwood(4) Rae & Cockrell
(5) Sonck et al.
(6) Appoldt & Bennett(6) Appoldt & Bennett(7) Pearson et al.
(7) Pearson et al.
(7) Pearson et al.
(8) Burgess and Moore(9) Chino et al.
(10) Steege et al.(11) Steege et al.(12) Steege & Childress(13) Steege & Childress(14) Brennan & Childress(15) Silver-Thom(16) Silver-Thom & Childress(17) Silver-Thom et al.(18) Torres-Moreno et al.(19) Sanders(20) Sanders et al.(21) Sanders & Daly(22) Sanders et al.(23) Sanders & Daly
transducer protrusion
diaphragm transducer
(24) Appoldt et al.
dynamic loading
FSR (Tekscan) (25) Springer & Engsberg(26) Engsberg et al.(27) Houston et al.
sensors have recently been incorporated into a pros-thetic measuring system (26,27,31) . None of theseprosthetic investigations, however, discuss the calibra-tion of the sensors, their accuracy, or other issuespertaining to their use.
The Socket Fitting System (R .G. Rincoe andAssociates, Golden, CO) uses similar technology toenable investigation of prosthetic socket interface pres-sures (32) . This commercial system allows measurement
of the interface pressures without damage to theprosthetic socket . Although this system allows examina-tion of a greater portion of the interface during aparticular gait trial than prior research investigations,the accuracy and repeatability of these transducers havenot been documented . In addition, data acquisition viathis system is currently limited to three specificinstances in the gait cycle : heel strike, mid-stance, andtoe-off.
256
Journal of Rehabilitation Research and Development Vol . 33 No. 3 1996
Numerical AnalysesIn contrast to the experimental techniques, com-
puter models of the residual limb and prosthetic sockethave the potential to estimate the interface pressures forthe entire residuum, and indeed are not limited to theinterface, but also can provide information regarding thesubcutaneous stresses . Nola and Vistnes and Daniel etal. have found that initial pathological changes inpressure sore formation occur first in the muscledirectly overlying the bone, and then spread outwardtoward the skin (33,34) . Therefore, the subcutaneousstresses may be of importance in evaluating long-termprosthetic success . As current measuring techniquesdisrupt the very stress distribution that is of interest,these subcutaneous stresses are particularly difficult tomeasure in vivo.
Several groups have used computer models of theresidual limb to investigate the residual limb/prostheticsocket interface . Nissan used a simplified three-dimensional (3-D) biomechanical model of the short BKresidual limb to investigate the effects of load transmis-sion area, tibial geometry, and the role of the fibula onthe forces and moments exerted on the residual limb(35) . Winarski et al . attempted to correlate prosthesisloads to interface stresses at the patellar tendon andgastrocnemius areas as a function of the flexion/extension angular alignment for persons with BKamputation (36).
Many investigators have also used finite element(FE) modeling of the residual limb and the prostheticsocket of persons with lower limb amputation toinvestigate residual limb/prosthetic socket biomechan-ics, and to estimate the interface pressure distributions.In fact, during the past 10 years, computer models ofthe residual limb and prosthetic socket have primarilyinvolved FE analysis . For this reason, the FE methodand the various formulations that have been utilized inthese prosthetic analyses will be reviewed in detail.
Introduction to the FE MethodThe FE method is a modeling technique that allows
investigation of complex structures : those having com-plicated geometry and/or complex material properties.For simple structures, the solution to the problem canoften be obtained analytically . For complex structures,however, an exact solution is rarely possible.
The FE method is an approximate solution methodwhereby a complex structure is discretized, or divided,into a number of regularly shaped pieces, known aselements . Each element is defined by several nodes, or
points, the coordinates of which establish the geometryof the structure . The behavior of the complex structureis then approximated as the sum of the respectiveresponses of each of the regularly shaped elements.
FE investigations may include structural (i .e ., stressanalysis, deformation analysis, fatigue analysis, crackpropagation), heat transfer, fluid flow, and/or electro-magnetic analysis . In biomedical engineering, structuralanalyses have been the most common . These investiga-tions have included stress analysis of prosthetic im-plants, bone remodeling, and current flow through theheart.
Overview of FE Modeling Formulations for theLower Limb Residual Limb
Complete FE model description requires the geom-etry, material properties, load state, and boundaryconditions of the model (Figure 1) . As prosthetic FEmodels have been based on various sources of suchinformation, and various formulations, each of thesefacets of FE model description will be discussed withrespect to past, present, and future models of theresidual limb/prosthetic socket system . Many optionsexist, and the selection of specific formulations isdependent upon and/or influenced by the underlyingpurpose of the investigation.
Terms identified by asterisks in the followingdiscussion are defined in the Glossary.
Figure 1.Sample FE model of a BK residual limb and prosthetic socket,illustrating key parameters in the model : limb geometry as definedby the nodal coordinates and the element connectivity, materialproperty description for all elements, boundary conditions andloading.
internal bone elements
the model iscomposed of8 noded solidhexahedronelements eachassigned material andgeometric properties
257
SILVER-THORN et al . Review of Prosthetic Stress Investigations
Model GeometryType of Analysis . FE analysis of the lower residual
limb has generally involved either two-dimensional(2-D, axisymmetric*) or 3-D models . Although neitherthe BK nor AK residual limbs are axisymmetric, 2-Dmodeling enables preliminary analysis that may indicatethe relative importance of certain modeling parameters.However, 2-D analysis has limited applicability infurthering prosthetic socket design, as the axisymmetricgeometry approximation limits application of realisticprosthetic socket rectifications . More complex 3-Danalyses usually follow 2-D analysis . These modelstypically require additional memory and computerrun-time due to the more complex representation oflimb geometry and the increased number of degrees of
freedom*, but 3-D analysis enables more accuraterepresentation of residual limb geometry and prostheticsocket shape.
Whether performing 2- or 3-D analysis, care mustbe taken to avoid warped* and skewed* elements*, andhigh aspect ratios*, all of which may contribute tocomputational errors in the analyses . Both linear andhigher order elements have been used in these FEmodels.
Source of Geometric Data
Data describing the geometry of the residual limbhave been based on the literature and non-contact and/orcontact methods such that either surface geometry only,or surface and internal (i .e ., bone) geometry wasobtained.
Literature-based anthropometric data is generic andnot individually specific . Such data can be scaled forparticular individuals . This data source will likely misssubtle variations in residual limb geometry that mayhave a significant role in prosthetic fit . Thus,anthropometric models may be most appropriate forparametric studies and/or qualitative analysis.
Non-contact methods of shape sensing have in-cluded traditional computer-aided design, computer-aided manufacture (CAD-CAM) optical surface meth-ods (laser scanners and Moire contours) as well asvolumetric imaging methods : X-ray, computed axialtomography (CAT or CT scans), magnetic resonanceimaging (MRI), and ultrasonic methods as shown inTable 2 (37-46) . These volumetric imaging techniques,which quantify both the internal and surface geometryof the residual limb, have been the primary non-contactmethod used in FE limb models . Ultrasound, CT, andespecially MRI data allow identification of individual
structures within the soft tissue bulk, and may enablemore complex representation of residual limb softtissues.
Imaging data, however, may also be subject toseveral limitations.
1. Imaging data is subject specific, and thus must beobtained and digitized for each individual . Thus,imaging techniques, and the subsequent digitiza-tion required for model incorporation, may bemore time-consuming and expensive than otherapproaches.
2. CT and X-ray techniques expose the subject toionizing radiation.
3. X-ray data is 2-D (planar projection of 3-D data);a minimum of two views are necessary for 3-Dapproximation of limb geometry . 3-D extrapola-tion of this data may be subject to large errors.
4. Identification of the prosthetic socket/residuallimb/bone boundaries in X-ray, CT, and/or MRIscans may require sophisticated image processingtechniques.
5. Many of these imaging techniques require that thepatient be supine or prone during data acquisition.As a result, the tissue geometry obtained may bedependent upon body position within the gravita-tional field.
In contrast to these non-contact methods, contactmethods are an atypical source of FE geometry forresidual limb models, but they are the most commonshape-sensing techniques used in CAD-CAM in pros-thetics today. In general, contact methods of shapesensing are susceptible to errors due to soft tissuedeformation associated with casting and/or calipermeasurements . These methods include : mechanicaldigitization of a prosthetic wrap cast (UCL CASD,Seattle ShapeMaker), reference model selection, andscaling based upon antero-posterior and medial-lateralmeasurements via calipers and circumferential measure-ments using a tape measure (MERU, University ofBritish Columbia).
Material PropertiesThe material properties of all of the structures
included in the FE model must be estimated andassigned to the respective elements . The source of theseestimates has been the literature, materials testing,and/or in vivo or in vitro studies . In general, thesematerials have been modeled as linear and nonlinearelastic isotropic* materials. In addition, both small and
258
Journal of Rehabilitation Research and Development Vol . 33 No . 3 1996
Table 2.
Summary of residual limb and/or prosthetic socket models reported in the literature.
•Geometry Loading
Material PropertiesMaxPRS
KPa
Bone o Soft Tissue o Socket• liner x pylon• skin/fat x muscle + canilage
type
E1
KPaReference Kind # VID Code Size type
1 SRC type
SRC type
E
vMPa t
type
E
v
SRCMPa
(13,14)
(13,14)
ANL
FND
I BK
2-BK
DL
PRS
NO
GIFTSMARCMARL
128017
ND
1976 ND578 ELI20 EL
3D SD
CT
3D SD
CT36
. LD
DL
STC
FP
STC
IDL
8
n/s
n s
8 ND
1 .6HX
8 ND
l
1.6)HX
.
•
D HX8 ND HX
790
I
49 END.49
LIT
• 8 ND HX
.38
= n/s
EXP:
EE
.38
F nJa i EXP
300
40
FND IDL NO 13362ND11676 EL
3D SD
CT !I LIT
QSI
LIT 8 ND
t 10 - 1 ?8HX
1500o 1 8 ND CD
2.6 - 96+
8 ND CD
100.499 1 LIT.499
IDL8 ND HX 1-2100 i .as
LIT8 ND HX
7000 i 28 # LIT200
Reynolds(53)
END II®
NO
970 EL
1458 ND972 EL500 DF
AX
DL
3D SD
XRY&
MDG
CAD/ID L
DSP CAD!DL
DSP BC
n/a 1 n/a
RGD BC , n/a
n la
o
20 ND
170AX
io
8 ND HX x 50 - 145
.45 : LND
AS IND
o SLIP, RGD
n/a i n/a
CAD& DSP BC
•DSP BC na CAD
370
Silver-Thorn(16,17)
END 2221 ND1688 EL
3D SD
CT /L1T
DSP CADSTC
FPRGD BC 1 n/a
n/ak
11
0
8 ND FIX)
.6 - 176
3
.45
END!LIT
0•
1
8 ND HX
1500
.30
END( 8 ND HX
.38
.45
LIT
IMMIMMiAliNEM 2400 ND2000 EL
3D SD
LDL
IT/•
RGD BC
n/a
n/a o
8 ND CD) 220 - 1802 .499
.45- LIT1
o• 1
.4530
LIT83 ND HH
XX
51-45K
.1ND
, .-
.
1
LII'1700
Quesada [DL NY) SAP80 636 ND 3D SD MDG STC 1 IDL 1 ~It
10 o . 2 ND ES 3
247 - 0 4 ND SH
14(8)0 ¢ .13
LIT58 655 EL HS
LIT 2490Sanders PRS 795 ND 3D SD
MRI QSI
SG RGD BC
n/a
n/a o t 8 ND HX
3.5-6 .9 .49 ; LIT o 4 ETD QD
n/s
n/s
n/s 90(20,24) ®® & 840EL 1 x j 8 N D HX)
131 .49 i LIT 8 ND ID'.
1 .8
.39
EXPSHR 1 l a beam
n/s
nls
?
n/sKrouskop END NO 3D SD
ITI LIT!
IDL 8 ND
n/s
nls o3
8 ND HX t
n/s n/s E USD o not
n/a
n/a
n/a n/a(61 Mill 1 USD HX
E I modeled , c-(62) RE* ANSYS 3D SD MDG PRS ` IDL RGD BC
n/a o . 8 ND HX t 53 - 141 n/s i LSD fl PRS BC
n/a
n/a 1 DL n/a63) ESEIIIIIWIEWIll ~ 3D SD , MDG PRS
DL RGD BC' n/a 1 n/a o ; 8 ND HX 1
5 - 10 nls : USD o PRS BC
n/a
n/a 2 IDLSeguchi 683 3D SD
CT IDL o65)
~m®®®®®~
34569
82673
NDEL
3D SD
CTIDL
/ NB IDL 8 ND
1 .6 j 28HX
10 1 8 ND CD 3
60 .49
LIT o DSP BC
nla
n!a
CAD 150
Torres- 1962 ND 3DLD
MRI DSP
EXP HX
n/s
n/s o
6 ND IN
27-146 .4999 i IND o 97ANL ABA-Moreno
®®QUS 628E L & x 1 6 ND IN
n!s(19,49) 7884DF PRS
IEWFND
® ®600AX
EL CT®RGD BC
n/a
n/a o 64 ND AXND HX
c RGD BC
Vannah END DL NO 108AX AX
DL DSP
DL RGD BC
n/a
n/a o i 4 ND IN
20 .7 .45 -
[ LN'D o not(47,48) 1
€ , 50
: modeled
KEY : AK=above-knee amputee, ANL=analytical, AX=axisymtnetric model, BC=boundars condition, BK=below-knee amputee, CAD=computer aided design, CD=constant dilitation, CT=computeraided tomography, DF=degrees of freedom, DSP=displacement, E=-Young's modulus, EF=elastic foundation. ES=elastic spring elements, EXP=experimental methods, FND=fundamental, FP=forceplate, HS=heel strike, HX=solid hexahedron element, IDL=idealized, IN=incompressible, IND=indentor tests, LD=large displacement, LIT=literature, MDG=mechanical digitization,MRI=magnetic resonance imaging, ND=node, NL=nonlinear, PRM=parametric, PRS=pressure,QD=quadrilateral, QSI=quasi-static, RGD=rigid, SD=small deformations, SG=strain gage, SHL=shellelement, SHR=shear, SLIP=slip allowed at interface, STC=static load, USD=ultrasounic method, v=Poisson's ratio, XRY=X-ray, n/a=not applicable, n/s=not stated.
large displacement* formulations have been attempted.
Such complexity may be required for soft tissues and/or
prosthetic materials that undergo significant deformation
under load . Variations in FE material property descrip-
tion will be discussed with respect to the specific
structures/materials commonly included in residual
limb/prosthetic socket models.
Prosthetic Socket and Liner
The prosthetic socket is typically composed of a
relatively stiff material, such as polypropylene or
polyester, and is approximately 3 to 6 mm thick. The
prosthetic socket has been explicitly modeled as:
• elements
• elastic foundations or springs
• a rigid kinematic boundary .
Modeling the prosthetic socket as elements re-
quires material property description . For example, if
the prosthetic socket is assumed to be a linearly
elastic isotropic homogeneous* material, material prop-
erty description includes specification of Young's modu-
lus* and Poisson's ratio* . Elastic foundations and
springs require material property description in terms
of force/area or force/length, respectively . The use
of elastic foundations and springs reduces the num-
ber of unknowns in the model (i .e., there are fewer
elements/nodes), but is limited in that the stress
distribution in the prosthetic socket cannot be calcu-
lated. In addition, application of various prosthetic
rectification schemes may be difficult to simulate.
Finally, the prosthetic socket has been modeled as
a rigid boundary via prescribed nodal displacements.
This formulation allows application of prosthetic rectifi-
259
SILVER-THORN et al . Review of Prosthetic Stress Investigations
cation schemes, but again neglects internal socketstresses.
The FE model may also include a soft liner, whichhas been similarly modeled as elements, elastic founda-tions, andlor linear springs.
Bone (Hard Tissue)
The bony structures of the residual limb also needto be incorporated in the FE model. In general, bone hasbeen modeled in one of two ways, either as elementswith properties estimated by literature values cited forcompact andlor cancellous bone, or as a fixed internalboundary, since bone is several orders of magnitudestiffer than soft tissue . By representing the bonystructures of the residual limb with elements, FEanalysis can be used to obtain estimates of bonystresses, motion of the bony structures with respect toone another, and bone motion within the soft tissuebulk. The latter formulation, a kinematic boundary, isless informative but results in a model with fewerdegrees of freedom, thus requiring less computermemory.
Soft Tissue
The soft tissue (muscle, skin, fat, ligaments, andtendons) of the residual limb is neither homogeneousnor isotropic . However, incorporation of homogeneousand isotropic material property assumptions greatlyreduces the complexity of the model . Consequently,although many of the investigators in this field havespeculated that precise representation of the tissueproperties is critical to model accuracy, the soft tissuehas typically been assumed to be an elastic, isotropic,homogeneous material (14,16,17,19,47) . Properties ofsoft tissue have been estimated from literature data, orexperimentally evaluated using in vivo indentor studiesof the residual limb (11,16,47—54).
The use of ultrasound, CT, or especially MRI dataprovides the capability of explicitly modeling theindividual structures of fat, muscle, and skin . However,if the model is to include such detail, these structuresmust be identified as specific elements, and the materialproperties of these respective elements must be esti-mated and assigned . Current models have yet to includethese structural complexities to any great degree.
An additional complexity of bulk soft tissue is thatit is believed to behave as a nearly incompressible*
material. This near incompressibility has been approxi-mated by several methods :
L
linear elastic formulations with Poisson's ratio, v,approaching 0 .5
2. constant dilatation* formulations in which softtissue is assumed to be incompressible on anelement basis
3. nonlinear elastomeric* formulations that utilizevarious forms of the strain energy equation(Mooney or Jamus-Green-Simpson formulations).
The first of these formulations, a linear elasticrepresentation of bulk soft tissue, may result in artifi-cially stiff behavior due to limits of displacementmethod FE analysis. To avoid this anomaly, reducedintegration elements may be used (55) . The constantdilatation formulation requires the introduction of apressure parameter which is equivalent to hydrostaticpressure in the limit of incompressibility (55) . Thisadditional parameter, which can be thought of as aLagrange multiplier, is incorporated as an additionalnode per element . Therefore, this method results inincreased complexity of the FE model, and requiresadditional computer memory due to the increase in thenumber of degrees of freedom per element . Theelastomeric formulations, defined by the strain energyequation, involve functions of the strain invariants . Thismethod has seen limited use in FE limb models, as theconstant coefficients in the strain energy equations arenot known, and the relationship between these coeffi-cients and the elastic moduli is not explicit (14,47,48).Most FE limb models have incorporated linear elasticformulations, 0 .45<v<0.49, due to the simplicity of thisformulation.
A final complexity of bulk soft tissue is itsviscoelasticity and dependence on load history . Bulksoft tissue, like its individual components, is believed tobe a viscoelastic material . Thus, its behavior, orresponse to loading, is dependent on the load rate andthe preconditioning of the tissue . Bulk soft tissue isexpected to demonstrate hysteresis upon loading/unloading, and exhibit stress relaxation and creep. Thepreconditioning of such tissues illustrates the depen-dence of the tissues' current loading response to priorloading . As yet, these phenomena have not beenincorporated in residual limb FE models.
Load StateThe load state of these FE residual limb models
has been either static or pseudo-static . In general, prioranalyses have concentrated on static loading, or stance,and pseudo-static approximations of gait . These load
260
Journal of Rehabilitation Research and Development Vol . 33 No. 3 1996
states have typically been based on data obtained fromforce plates or other force transducers during studies ofamputee gait and/or balance, subject to alignmentvariations. The loading has been applied to theelements/nodes of the bone, although models whichrepresent the bony structures of the residual limb as afixed boundary have applied load to the distal nodes ofthe prosthetic socket.
For a linear model, superposition is valid . There-fore, the analysis of various load states for linear modelsmay be approximated as the sum of prior analyses ofsimplified load components . For example, the initialstress state due to prosthetic socket rectification may besummed with the results of static loading.
Dynamic analysis introduces many additional pa-rameters, such as time-load increments and numericalintegration methods, that may greatly complicate theanalysis and influence model stability . Such analysismay also necessitate the inclusion of viscoelastic*phenomenon. Currently, all dynamic loading has con-sisted of pseudo-static analysis, which may be approxi-mated as a series of superposed load steps.
Boundary ConditionsThe boundary conditions of the reported residual
b models range from simple to complex . Thesimplest formulations assume total contact between thebone/soft tissue, soft tissue/prosthetic liner, and pros-thetic liner/socket interfaces . More complex formula-tions allow time-varying boundary conditions.
The preceding section on prosthetic sockets indi-cated that the socket interface has been approximated asan elastic foundation or as elastic springs . Theseinvestigations of the residual limb/prosthetic socketinterface only report normal stresses (pressures) at theinterface, and ignore, due to the total contact condition,the shear stresses . While deformation of the limb isallowed along the socket wall, slip cannot be computedwith these FE models (i .e ., the models assume infinitefriction).
In addition, tensile normal stresses may develop atthe residual limb surface in the FE models . As tensioncannot exist to any great extent between a limb and asocket, the existence of tension in FE models may becounter-intuitive . Therefore, attempts have been madeto reduce these tensile stresses through various iterativemeans.
Finally, more complicated formulations, whichmodel the structures as deformable and rigid bodies,allow the influence of contact and slip/friction to be
investigated. However, such formulations increase thecomplexity of the model, as every increment must beanalyzed to determine whether contact has been made,or broken, between the respective bodies. The ability ofcommercial FE software to perform 3-D contact analy-sis is a relatively recent development, and residual limbmodels are only now being developed that incorporatethis method of analysis.
Specific Residual Limb FEMsA number of investigations of the residual limb/
prosthetic socket system have been carried out using FEanalysis methods, and these approaches are reviewed inthe following paragraphs . Special attention is given tothe FE formulation parameters, and the potential effectsthese parameters have on model output . The formulationparameters of the respective models are summarized inTable 2.
Steege et al . were the first to model the residuallimb and prosthetic socket system for persons with BKamputation (11,12) . Their work included both FEanalysis of the residual limb and prosthetic socket, andexperimental verification of local interface pressures fortwo individuals with BK amputation . The internal andexternal geometry of the residual limbs was based ontransverse CT scans of the residual limb and unrectifiedprosthetic socket . The soft tissue was assumed to be ahomogeneous, linearly elastic, isotropic material, andthe value of Young's modulus (E=60 kPa) was based onthe mean results of in vivo indentor studies of the softtissues of the residual limb at several locations . Thematerial properties of the bone, cartilage, and patellartendon were based on the literature . The load state forthese analyses and for the pressure measurements wasstatic double support stance. In the initial FE analyses,the range of estimated pressures was 0—105 kPa . Morerecently, Steege et al . have reported using FE analysisto design BK prosthetic sockets, as well as performingquasi-static analysis based upon dynamic gait data(56,57)
Reynolds (53) and Reynolds and Lord (54) alsoattempted to estimate BK prosthetic interface pressures.Initial parametric analyses, investigating the effects offriction, material properties, and socket design, wereconducted for a 2-D, axisymmetric FE approximation ofthe BK residual limb . Verification of the interfacepressure estimates was performed using a physicalmodel of this axisymmetric approximation of theresidual limb. These analyses were of limited clinicalvalue, since the actual geometry of the residual limb
261
SILVER-THORN et al . Review of Prosthetic Stress Investigations
was poorly represented, and the imposed socket rectifi-cation schemes were highly artificial . Reynolds thendeveloped and analyzed a 3-D model of the BK residuallimb based on radiographic data. The properties of thebulk soft tissue were assumed to be linear, with localYoung's moduli based on in vivo indentor tests at foursites on the residual limb (50–145 kPa) . Analyses wereperformed for various socket rectification schemes andfor various material property approximations of bulksoft tissue for static, double support loading . Pressuresranged from 0 to 200 kPa for the nominal limb model.No verification/validation of the 3-D model was per-formed.
Sanders (20) and Sanders and Daly (22,24) investi-gated BK interface pressures using both 3-D linear FEanalysis of the residual limb and prosthetic socket, andexperimental measurement of interface stresses . Sand-ers' work differed from previous research in that: 1)stress measurements included both normal (pressure)and shear stress, 2) the load state for the FE model wasdynamic (i .e ., quasi-static representations of gait), 3) theresidual limb geometry was based upon MR images,and 4) the FE model approximated bulk soft tissue as acombination of skin/fat and muscle.
Quesada and Skinner used FE analysis of a PTBprosthesis to investigate variations in prosthesis designon the interface stress distribution at heel strike (58).These models approximated the bulk soft tissue of theresidual limb as parallel (skin) and perpendicular(compressive tissue) linear springs attached to thesocket wall . The normal stresses estimated with thismodel ranged from 0 to 961 kPa ; the shear stressesranged from 0 to 463 kPa . The stresses estimated at thedistal anterior end of the residual limb/socket (961 kPanormal stress, 463 kPa shear stress) were considerablyhigher than the stresses predicted for the remainder ofthe limb/socket . No verification of the model has beenreported.
Silver-Thorn (16) and Silver-Thom and Childress(17,59) created a FE model of the BK residual limb andprosthetic socket to investigate the effects of parametervariations on the interface pressure distribution duringstatic stance . The geometry of both the residual limband prosthetic socket in this model were approximatedby standard geometric shapes, the size and selection ofwhich were based on available anthropometric data . Thefemur, tibia, fibula, and patella, as well as the articularcartilage of the knee joint, were modeled as a fixedinternal boundary . For the parametric studies, the softtissue, assumed to be isotropic, linearly elastic, and
homogeneous, was assigned a Young's modulus of 60kPa (Poisson's ratio=0 .45) . The prosthetic socket andliner were modeled as elements, the material propertiesof which were based on the literature . The load state"double support" stance, was applied to the distal endof the socket . Parametric analyses were performed toinvestigate the effects of variations in : 1) the materialproperties of the residual limb, prosthetic liner, andprosthetic socket; 2) the internal and external geometryof the residual limb; and 3) the prosthetic socketrectification scheme . Analyses were conducted for aperson with BK amputation for both unrectified andPTB rectified socket designs . The ability of this genericgeometric FE model to estimate prosthetic interfacepressures was evaluated based upon comparison of theFE interface pressures (individually scaled models) tothat measured for three people with BK amputation in avariety of socket designs and static load/alignmentstates (16,17,56).
Krouskop et al . were the first to make use of theFE method as a CAD tool for AK prosthetic sockets(60–62). After evaluating the surface geometry of theresidual limb through a contact method using twodiametrically opposed contracting/retracting probes,ultrasound was used to obtain average local materialproperties . A generic FE model was then scaledappropriately for the subject's surface geometry, and thelocal material properties were assigned to respectivelinear elastic 3-D elements . A static loading function,based on average interface pressure profiles measuredfor subjects wearing comfortable quadrilateral-brim AKprostheses was imposed . The FE model was then usedto predict the shape of the loaded limb such that thedesired pressure profile would be obtained . This recti-fied socket geometry was then carved on a computernumerically carved (CNC) milling machine, and theproposed socket was subsequently vacuum formed.Krouskop reported the successful fitting of two personswith AK amputation using this methodology.
Research has also been conducted using FEanalysis to study the interface pressure distribution forpersons with AK amputation . The models developed byBrennan and Childress (15), Torres-Moreno et al . (19),and Mak et al . (63) are similar to the FE models for BKresidual limbs and sockets mentioned previously . Themodel developed by Seguchi et al . was novel (64).Seguchi avoided characterization of the mechanicalproperties of bulk soft tissue by modeling only theacrylic socket . As this problem is underde fined, thecomplementary energy criterion was used to search for
262
Journal of Rehabilitation Research and Development Vol . 33 No. 3 1996
the most plausible interface pressure distribution . TheFE model was based on transverse CT scans of thesocket, and consisted of thin quadrilateral shell ele-ments . The static response of the socket was investi-gated for two hypothetical load cases : 1) uniformcontact pressure along the interior surface of the socket,and 2) weight fully supported at the ischial seat.Experimental verification of both the model and themethod consisted of circumferential strain measure-ments during point loading on the anterior and posteriorwalls of the socket. The clinical value of such a modelis questionable, however, as the model ignores residuallimb geometry and bone-soft tissue interactions.
SUMMARY
Many of the interface pressure measurement tech-niques involve research applications and methodologiesthat are not appropriate for routine clinical use . How-ever, the development of commercial systems using thinforce sensitive resistive materials to measure interfacepressures/stresses for both seating systems and pros-thetic sockets may provide a diagnostic tool that can bereadily incorporated into prosthetic fitting . These sys-tems have clinical potential and may facilitate creationof prosthetic pressure databases such that interfacepressures, whether measured in research settings, clini-cal settings, or estimated with computer models, may beproperly interpreted.
We believe that the two greatest limitations incurrent modeling efforts involve the representation oftissue properties across the entire limb and the interfacecondition between the residual limb and prosthesis . Theability of current FE models to estimate prostheticinterface stresses, while in some cases performingreasonably well, has not been highly accurate on thewhole . Nevertheless, the methodology has distinctpromise and potential, and the trend is toward improvedaccuracy. Advances in FE software, enabling nonlinearelastomeric formulations of bulk soft tissue, contactanalysis, and dynamic analysis may help to addresssome of the current limitations of the models . Inaddition, the advances in computer hardware makeapplication of these modeling complexities feasible.Thus, the tools needed to advance the FE models areavailable, and continued progress appears likely . Corre-sponding advances in the pressure transducer technol-ogy will help to validate the computer models, and alsofacilitate interpretation of the results of the analyses .
Applications and Utility of Prosthetic InterfaceStress Studies
Clinical measurement of the prosthetic interfacepressures has the potential to provide quantitative,objective information to assist in the evaluation ofprosthetic fit . In addition, the effects of prostheticalignment and componentry on the interface stressdistribution can be determined. Routine use of suchsystems/devices will help to establish a database so thatthe significance of the results might be established andassist in the interpretation of the results.
Regardless of the assumptions and the simplifica-tions of various computer models, numerical analysis ofthe residual limb and socket offer several advantagesover experimental measurements in the estimation ofprosthetic interface pressures . For example, the use ofcomputer models, as opposed to experimental analysis,allows examination of the entire residual limb/prostheticsocket interface . In addition, prospective socket designs,characterized by material modifications and/or alterna-tive socket rectification schemes may be investigatedprior to socket manufacture. In fact, hypotheticaldesigns that cannot be fabricated due to currenttechnological limitations (i .e ., material constraints) maybe investigated. Note that computer models necessitateexperimental verification to establish model validity.
Currently, it is not possible to perform clinicalparametric studies of the prosthetic socket due todifficulties in the repeatability of test procedures, cost,and time constraints . Computer models are not subjectto these limitations, and provide the potential forextensive parametric analysis.
These advantages of numerical analysis might alsobe utilized to enhance the education and training ofprosthetists. Graphical results, still or animated, gener-ated from computer models investigating prostheticalignment and socket shape may be incorporated intoprosthetic training software . Such software would allowthe prosthetist to visualize the stresses resulting fromhis/her design efforts.
Future TrendsThe recent availability of commercial systems to
measure and record prosthetic interface pressures (andpossibly shears) should provide additional objectivitywhen evaluating prosthetic fit . However, these systemsand their respective transducers must be thoroughlycharacterized so that results for various individuals andsystems may be interpreted appropriately . These devicesmay make it possible to create databases of interface
263
SILVER-THORN et al . Review of Prosthetic Stress Investigations
pressures (and shears) experienced during amputeestance/gait . Such information will be useful in evaluat-ing prosthetic fit, and in designing prosthetic sockets,whether the design involves CAD-CAM or conventionaltechniques.
FE analysis of the lower residual limb andprosthetic socket system will likely continue in thefuture . The FE method has proven itself as an extremelyuseful general engineering tool, and while its applica-tion to limb prosthetics has not been as straightforwardas initially hoped, the FE method holds promise forparametric analysis of prosthetic systems . These com-puter models offer distinct advantages over experimen-tal measurements, in that stress information can beobtained for the entire prosthetic interface, and that thesubcutaneous stress distribution may also be investi-gated . In addition, parametric analysis, aimed at improv-ing the understanding of the residual limb prostheticsystem, is only feasible via computer analysis.
FE models have potential applicability in CAD ofprosthetic sockets . Current prosthetic CAD systemsallow the prosthetist to replicate on the computer whathe has always done with his hands, that is, changegeometry in an attempt to control force distribution inthe residual limb. We foresee the FE technique beingincorporated as the "engine" of future CAD (orperhaps more appropriately computer aided engineering:CAE) programs . Similar to current methods, prosthetistswill use the computer mouse to identify areas on agraphical rendering of a residual limb . Instead ofmaking shape changes to the surface of the limb, theywill prescribe pressure tolerance information to themodel. This optimal load information will be the inputto the FE-based CAE design program ; the output will bethe computed shape of the new socket . In this manner,the prosthetic design will be based directly on thecontrol parameter, namely force or pressure.
Finally, future developments in CAD of prostheticsockets are also likely to be influenced by alternativeshape sensing methodology, such as ultrasound, castdigitizers, and spiral CT scanning, that will enabletimely evaluation of residual limb geometry and/ormaterial properties.
Future FE models will likely vary from thosesummarized in this paper . Advances in computertechnology (faster processors and larger memory) allowanalysis of larger, more complex models . In addition,developments in FE software enable incorporation ofnonlinear material properties, contact analysis, and truedynamic analysis . The development of commercial or
custom software to process imaging data and automateFE mesh generation, similar to packages for customprosthetic hip design, should decrease the time neededto create these models . The incorporation of thesefeatures will likely have a significant impact on theperformance of residual limb/prosthetic socket models,and the future utility of such models in prostheticdesign.
GLOSSARY
Aspect Ratio . Ratio of the longest to shortest elementdimensions ; to minimize computational errors, onegenerally wants element aspect ratios less than 3for linear elements and less than 10 for quadraticelements.
Axisymmetric Model . Structure to be modeled is abody of revolution ; the model geometry, materialproperties, and boundary conditions can be mod-eled in cylindrical coordinates (r,O,z), and are suchthat all are independent of O. Such criteria enable3-D structures/problems to be analyzed in 2-D,thereby reducing the complexity of the model.
Degrees of Freedom . Number of nodal displacementsand rotations allowed in a particular problem . Forexample, for 2-D problems, each node can move inthe x- and y-directions; thus, there are two degreesof freedom per node.
Dilatation. Volume change; for constant dilatationproblems, no volume change is allowed (i .e .,structures are incompressible) . The constant dilata-tion constraint is enforced on an element basis viaan additional node.
Elastomer . "Rubberlike" polymeric material that cansustain very large elastic deformations ; the behav-ior is usually approximated by various formula-tions of the strain energy equation.
Element. There are many types of element formula-tions, including linear elements in which theelement sides remain straight and higher orderelements in which the element sides may becurved. The order of the element affects theelement shape and the corresponding elementstrain and stress. Examples of linear elementsinclude 3-node triangles and 4-node quadrilaterals(2-D) and 4-node tetrahedrons and 8-node bricks(3-D). Quadratic elements include 6-node trianglesand 8-node quadrilaterals (2-D) and 10-node tetra-hedrons and 20-node bricks (3-D) .
264
Journal of Rehabilitation Research and Development Vol . 33 No. 3 1996
12.
13.
14.
15.
16.
17.
18.
1 . Meier RH, Meeks ED, Herman RM . Stump-socket fit ofbelow-knee prostheses : comparison of three methods of
19.measurement. Arch Phys Med Rehabil 1973 :54:553-8.
Homogeneous . Material which is uniform in structureor composition.
Incompressible . Material whose volume cannot bereduced due to loading.
Isotropic . Material whose elastic properties are identi-cal in all directions.
Large Displacement . For a large displacement formu-lation, the dependence of the stress-strain relation-ship upon the current state of strain (or displace-ment) is explicitly modeled . This formulation issynonymous with nonlinear geometric modeling.
Linear FE Model . In linear FE models, the elementmaterial properties and model geometry are linear.In contrast, nonlinear FE analysis may includenonlinear material properties, geometry (bucklingand/or large displacement) and/or boundary condi-tions.
Poisson's Ratio (v). Ratio of the transverse to longi-tudinal strain under uniaxial tension.
Skewness. Variation of element vertex angles from90° for quadrilaterals or from 60° for triangularelements. To minimize computational errors, onewants to minimize excess skewness or elementdistortion.
Viscoelasticity . The behavior of a viscoelastic mate-rial is a function of the load rate . Viscoelasticmaterials exhibit hysteresis, stress relaxation,and/or creep.
Warpage . Relevant to 3-D analysis when a node on asingle face of a solid (or plate or shell element)deviates from a single plane. To minimize compu-tational errors, excess warpage or element distor-tion should be minimized.
Young's Modulus (E) . Measure of material propertystiffness . Young's modulus is the slope of thelinear portion of the stress-strain curve for elasticmaterials.
ACKNOWLEDGMENTS
The authors would like to acknowledge the coop-eration of the Lakeside VA Medical Center, Chicago, ILand the Clement J . Zablocki VA Medical Center,Milwaukee, WI .
2. VanPijkeren T, Naeff M . Kwee HH. A new method for themeasurement of normal pressure between amputation residuallimb and socket . Bull Prosthet Res 1980 :10-33:31-4.
3. Isherwood PA. Simultaneous PTB socket pressures and forceplate values . Biomechanical Research and Development UnitReport, London, 1978 :45-9.
4. Rae JW, Cockrell JL . Interface pressure and stress distributionin prosthetic fitting . Bull Prosthet Res 1971 :10-16 :64-111.
5. Sonck WA, Cockrell JL, Koepke GH. Effect of liner materialson interface pressure in below-knee amputees . Arch Phys MedRehabil 1970 :51 :666-9.
6. Williams RB, Porter D, Roberts VC, Regan JF . Triaxial forcetransducer for investigating stresses at the stump/socketinterface . Med Biol Eng Comput 1992 :30 :89-96.
7. Appoldt FA, Bennett L . A preliminary report on dynamicsocket pressures . Bull Prosthet Res 1967 :10-8:20-55.
8. Pearson JR, Holmgren G, March L, Oberg K . Pressure incritical regions of below-knee prostheses . Bull Prosthet Res1973 :10-19:53-76.
9. Burgess EM, Moore AJ . A study of interface pressures in thebelow-knee prosthesis . Bull Prosthet Res 1971 :10-16 :58-70.
10. Chino N, Pearson JR, Cockrell JL, Mikishko HA, KoepkeGH. Negative pressures during swing phase in below-kneeprostheses with rubber sleeve suspension . Arch Phys MedRehabil 1975 :56:22-6.Steege JW, Schnur DS, VanVorhis RL, Rovick JS . Finiteelement analysis as a method of pressure prediction at thebelow-knee socket interface . In: Proceedings of the 10thAnnual RESNA Conference, 1987, San Jose, CA . Washing-ton, DC: RESNA Press, 1987 :814-6.Steege JW, Schnur DS, Childress DS . Prediction of pressurein the below-knee socket interface by finite element analysis.In : ASME Symposium on the Biomechanics of Normal andPathological Gait, 1987:39-44.Steege JW, Childress DS . Finite element prediction ofpressure at the below-knee socket interface . In: Report ofISPO Workshop on CAD/CAM in Prosthetics and Orthotics,1988:71-82.Steege JW, Childress DS . Finite element modeling of thebelow-knee socket and limb : phase II . In: Modeling andcontrol issues in biomechanical systems, ASME, 1988 :121-9.Brennan JM, Childress DS . Finite element and experimentalinvestigation of above-knee amputee limb/prosthesis systems:a comparative study . In : Proceedings of the ASME, advancesin bioengineering, 1991 :547-50.Silver-Thom MB. Prediction and experimental verification ofresidual limb/prosthetic socket interface pressures for below-knee amputees (Dissertation) . Chicago: Northwestern Univer-sity, 1991.Silver-Thorn MB, Childress DS . Use of a generic, geometricfinite element model of the below-knee residual limb andprosthetic socket to predict interface pressures . In : Proceed-ings of the 7th World Congress of ISPO, 1992 :272.Silver-Thom MB, Steege JW, Childress DS . Measurements ofbelow-knee residual limb/prosthetic socket interface pressures.In: Proceedings of the 7th World Congress of ISPO,1992 :280.Torres-Moreno R, Solomonidis SE, Jones D . Three-dimensional finite element analysis of the above-knee residual
REFERENCES
265
SILVER-THORN et al . Review of Prosthetic Stress Investigations
limb . In : Proceedings of the 7th World Congress of ISPO,
1992 :274.20. Sanders JE . Ambulation with a prosthetic limb : mechanical
stresses in amputated limb tissues (Dissertation) . Seattle:
University of Washington, 1991.21. Sanders JE, Daly CH, Burgess EM. Interface shear stresses
during ambulation with a below-knee prosthetic limb . J
Rehabil Res Dev 1992 :29(4) :1-8.
22. Sanders JE, Daly CH . Measurement of stresses in threeorthogonal directions at the residual limb prosthetic socketinterface . IEEE Trans Rehabil Eng 1993 :1 :79-85.
23. Sanders JE, Daly CH, Burgess EM . Clinical measurement ofnormal and shear stresses on a trans-tibial stump : characteris-tics of wave-form shapes during walking . Prosthet Orthot Int1993 :17 :38-48.
24. Sanders JE, Daly CH. Normal and shear stresses on a residuallimb in a prosthetic socket during ambulation : comparison offinite element results with experimental measurements. J
Rehabil Res Dev 1993 :30(2) :191-204.25. Appoldt FA, Bennett L, Contini R . Socket pressure as a
function of pressure transducer protrusion . Bull Prosthet Res
1969 :236-49.26. Springer MJN, Engsberg JR . Quantifying interface pressures
in below-knee amputee sockets . In : Proceedings of NACOB
II, 1992:401.27. Engsberg JR, Springer JN, Harder JA . Quantifying interface
pressures in below-knee amputee sockets . J Assoc ChildProsthet-Orthot Clin 1992 :37(3) :81-8.
28. Ferguson-Pell MW . Design criteria for the measurement ofpressure at body/support interface . Eng Med 1980 :9(4) :209-
14.29. Ferguson-Pell MW, Bell F, Evans JH. Interface pressure
sensors : existing devices, their suitability and limitations.In : Kennedy RM, Cowden JH, Scales JT, eds . Bedsore
biomechanics . Baltimore : University Park Press, 1976 :189-
97.30. Lord M, Hosein R . Pressure redistribution by moulded inserts.
In : Proceedings of the 7th World Congress of ISPO,
1992:279.31. Houston VL, Mason CP, LaBlanc KP, Beattie AC, Garbarini
MA, Lorenze E.J . Preliminary results with the DVA-Tekscanbelow-knee prosthetics socket/ residual limb stress measure-ment system. In : Proceedings of the 20th Annual Meeting &Scientific Symposium of the American Academy of Orthotistsand Prosthetists, 1994 :8-9.
32.
Socket Fitting System, RG Rincoe and Associates, Golden,CO.
33. Nola GT, Vistnes LM . Differential response of skin andmuscle in the experimental production of pressures sores.Plast Reconstr Surg 1980:66 :728.
34. Daniel RK, Priest, DL, Wheatley DC . Etiologic factors in
pressure sores : an experimental model . Arch Phys Med
Rehabil 1981 :62(5):728-33.35.
Nissan M. A simplified model for the short-below-knee
stump . J Biomech 1977 :10 : 651-8.36. Winarski DJ, Pearson JR, Cockrell JL . Prosthetic loads and
stump pressure correlations . In : Proceedings of the ASMEWinter Annual Meeting: advances in bioengineering,
1982 :29-32 .
37. Oberg K, Kofman J, Karisson A, Linstrom B, Sigblad G . TheCAPOD system : a Scandinavian system for prosthetic sockets.J Prosthet Orthot 1989 :1(3) :139-48.
38. Walsh NE, Lancaster JL, Faulkner VW, Rogers WE. Acomputerized system to manufacture prostheses for amputeesin developing countries . J Prosthet Orthot 1989 :1(3) :165-81.
39. Saunders CG. Reconstruction of anatomic shapes from moirecontourgraphs . In : Proceedings of SPIE Biostereometrics '82,1982 :361 :99-106.
40. Lee VS, Solomonidis SE, Spence WD . Magnetic imaging ofthe trans-femoral redisual limb . In: Proceedings of the 8thWorld Congress of ISPO, 1995 :127.
41. He P, Xue K, Bismar HA, Fu H, Murka P . Application ofultrasonic imaging technique to prosthetic socket design . In:Proceedings of the Annual International Conference of theIEEE Engineering in Medicine and Biology Society.1991 :3 :1849-50.
42. He P, Xue K, Murka P, O'Bleness F . A PC-based ultrasonicdata acquisition system for computer aided socket design . In:Proceedings of the 19th Annual Meeting & ScientificSymposium of the American Academy of Orthotists andProsthetists, 1993 :22.
43. Murka P, Voogt S . Ultrasound imaging system applied toprosthetic socket design . In : Proceedings of the 20th AnnualMeeting & Scientific Symposium of the American Academyof Orthotists and Prosthetists, 1994 :55.
44. Faulkner V, Walsh N, Gall N . A computerized ultrasoundshape sensing mechanism . Orthot Prosthet 1988:41(4) :57-65.
45.
Faulkner V, Walsh N, Gall N . Ultrasound as an aid toprosthetic socket design . J Rehabil Res Dev Progress Reports1986 :7.
46. Rogers B, Morimoto AK, Faulkner V, Kozlowski DM, WalshNE, Vick DD . Evaluation of a 3D ultrasound imaging systemfor BK prosthesis fabrication . In: Proceedings of the 8thWorld Congress of ISPO, 1995 :126.
47. Vannah WM . Indentor tests and finite element modelling ofbulk muscular tissue in vivo (Dissertation) . Chicago: North-western University, 1990.
48. Vannah WM, Childress DS. Modelling the mechanics ofnarrowly contained soft tissues : the effects of specification ofPoisson's Ratio . J Rehabil Res Dev 1993 :30(2) :205-9.
49. Torres-Moreno R, Solomonidis SE, Jones D . Geometricaland mechanical characteristics of the above-knee residuallimb. In: Proceedings of the 7th World Congress of ISPO,1992 :149.
50. Torres-Moreno R, Solomonidis SE . Jones, D . Load transfercharacteristics at the socket interface of above-knee amputees.In: Proceedings of the 7th World Congress of ISPO,1992 :151.
51. Mak AFT, Liu GHW, Lee SY . Biomechanical assessment ofbelow-knee residual limb tissue . J Rehabil Res Dev1994 :31(3) : 188-98.
52. Mak AFT, Zheng YP. Biomechanical assessment of stumptissues using a force transducer in series with an ultrasoundthickness gage . In : Proceedings of the 8th World Congress ofISPO, 1995 :127.
53. Reynolds D. Shape design and interface load analysis forbelow-knee prosthetic sockets (Dissertation) . London : Univer-sity College, 1988 .
266
Journal of Rehabilitation Research and Development Vol . 33 No . 3 1996
54.Reynolds DP, Lord M. Interface load analysis for computer-
60.aided design of below-knee prosthetic sockets . Med Biol EngComput 1992 :419-26.
55. Hughes TJR. The finite element method : linear static anddynamic finite element analysis . New Jersey : Prentice-Hall,Inc ., 1987 :185-309 .
61.56. Steege JW, Silver-Thorn MB, Childress DS . Design of
prosthetic sockets using finite element analysis . In : Proceed-ings of the 7th World Congress of ISPO, 1992 :273.
57.
Steege JW, Childress DS . Analysis of trans-tibial prosthetic
62.gait using the finite element technique . In : Proceedings of the21st Annual Meeting & Scientific Symposium of the Ameri-can Academy of Orthotists and Prosthetists, 1995 :13-4.
58.
Quesada P, Skinner HB . Analysis of a below-knee patellar
63.tendon-bearing prosthesis : a finite element study . J RehabilRes Dev 1991 :28(3) :1-12.
59.
Silver-Thorn MB, Childress DS . Sensitivity of below-knee
64.residual limb/prosthetic socket interface pressures to varia-tions in socket design . In: Proceedings of the 7th WorldCongress of ISPO, 1992 :148 .
Krouskop TA, Goode BL, Dougherty DR, Hemmen EH.Predicting the loaded shape of an amputees residual limb.In: Proceedings of the 8th Annual RESNA Conference,1985, Memphis, TN . Washington. DC: RESNA Press,1985 :225-7.Krouskop TA, Muilenburg AL, Dougherty DR , Winning-ham DJ . Computer-aided design of a prosthetic socketfor an above-knee amputee. J Rehabil Res Dev1987 :24(2) :31-8.Krouskop TA, Malinauskas M, Williams J, Barry PA,Muilenburg AL, Winningham DJ . A computerized method forthe design of above-knee prosthetic sockets . J Prosthet Orthot1989 : 1(3) : 13 1-8.Mak AFT, Yu YM, Hong ML Chan C . Finite element modelsfor analyses of stresses within above-knee stumps. In:Proceedings of the 7th World Congress of ISPO, 1992 :147.Seguchi Y, Tanaka M, Akazawa Y, Nakagawa A, Kitayama I.Finite element analysis and load identification of above-kneeprosthesis socket. ANSYS Conference Proceedings, 1989:31-44.